Google Pressure Wave: Do the Big Boys Feel It?

February 25, 2008

In 2004, I began work on The Google Legacy: How Google’s Interent Search Is Transforming Application Software. The study grew from a series of research projects I did starting in 2002. My long-time colleague, friend, and publisher — Harry Collier, Infonortics Ltd. in Tetbury, Glou. — suggested I gather together my various bits and pieces of information. We were not sure if a study going against the widely-held belief that Google was an online ad agency would find an audience.

The Google Legacy focused on Google’s digital “moving parts” — the sprockets and gears that operate out of sight for most. The study’s major finding was that Google set out to solve the common problems of performance and relevance in Web search. By design or happenstance, the “solution” was a next-generation application platform.

The emergence of this platform — what I called the Googleplex, a term borrowed from Google’s own jargon for its Mountain View headquarters — took years. Its main outlines were discernable by 2000. At the time of the initial public offering in 2004, today’s Googleplex was a reality. Work was not finished, of course, and probably never will be. The Googleplex is a digital organism, growing, learning, and morphing.

The hoo hah over Google’s unorthodox IPO, the swelling ad revenue, the secrecy of the company, and the alleged arrogance of Googlers (Google jargon for those good enough to become full-time employees) generated a smoke screen. Most analysts, pundits, and Google watchers saw the swirling fog, but when The Google Legacy appeared, few had tried to get a clearer view.

Google provided some tantalizing clues about what its plumbing was doing. Today, you can download Hadoop and experiment with an open source framework similar to Google’s combo of MapReduce and the Google File System. You can also experiment with Google’s “version” of MySQL. Of course, your tests only provide a partial glimpse of the Google’s innards. You need the entire complement of Google software and hardware to replicate Google.

Google also makes available a number of technical papers, instructional videos, lectures, and code samples. A great place to start your learning about Google technical innovations is here. If you have an interest in the prose of folks more comfortable with flashy math, you can read Google technical papers here. And, if you want to dig even more deeply into Google’s mysteries, you can navigate to the US Patent & Trademark Office and read the more than 250 documents available here. The chipper green and yellow interface is a metaphor for the nausea you may experience when trying to get this clunky, Bronze Age search system to work. But when it does, you will be rewarded with a window into what makes the Google machine work its magic.

The Google Legacy remains for some an unnerving look at Google. Even today, almost three years after The Google Legacy appeared, many people still perceive Google as an undisciplined start up, little more than a ersatz college campus. The 20-somethings make money by selling online advertising. I remember reading somewhere that a Microsoft executive called the Google “a one-trick pony”.

You have to admit. For a company that will be 10-years-old in a few months, a canny thinker like Steve Ballmer has perceived the company correctly. But why not ask this question, “Has Microsoft really understood Google?” A larger and more interesting question, “Have such companies as IBM, Oracle, Reed Elsevier, and Goldman Sachs grasped Google in its entirety?”

In this essay, I want to explore this question. My method will be to touch upon some of the information my research uncovered in writing the aforementioned The Google Legacy and my September 2007 study, Google Version 2.0. Then I want to paraphrase a letter shared with me by a colleague. This letter was a very nice “Dear John” epistle. In colloquial terms, a very large technology company “blew off” my colleague because the large technology company understood Google and didn’t need my colleague’s strategic counsel about Google’s enterprise software and service initiatives.

I want to close by considering one question, “If Microsoft is smart enough to generate more than $60 billion in revenue in 2007, why hasn’t Microsoft been clever enough to derail Google?” By correspondence, if Microsoft didn’t understand Google, can we be confident that other large companies have “figured out Google”?

Microsoft Should Stalk Other Prey, Says New York Times

Today is February 25, 2008, there’s still a “cloud of unknowing” around Google. One Sunday headline “Maybe Microsoft Should Stalk Different Prey” caught my eye. The article here, penned by Randall Stross includes this sentence:

Having exhausted its best ideas on how to deal with Google, Microsoft is now working its way down the list to dubious ones — like pursuing a hostile bid for Yahoo.

Now Microsoft has been scrutinizing Google for years. Microsoft has power, customers, and money. Microsoft has thousands of really smart people. Google — in strictly financial measures — is a dwarf to Microsoft’s steroid stallion. Yet I was learning from the outstanding, smart reporter Randall Stross revealing that the mouse (Google) has frightened the elephant (Microsoft). Furthermore, the elephant can’t step on the mouse. The elephant cannot move around the mouse. The elephant has to do the equivalent of betting the house and children’s college fund to have an chance to escape the mouse.

Microsoft seems to be faced with some stark choices.

SWAT

For me, the amusing part of this Sunday morning “revelation” is that by the time The Google Legacy appeared in 2005, Microsoft was between a rock and a hard place with regards to Google. One example from my 2005 study will help you understand my assertion.

Going Fast Cheaply

In the research for The Google Legacy, I read several dry technical papers about Google’s read speed on Google’s commodity storage devices. A “read speed” is a measure of how much data can be moved from a storage device to memory in one second. Your desktop computer can move megabytes a second pretty comfortably. To go faster, you need the type of engineering used to make a race car outperform your family sedan.

These papers, still available at “Papers Written by Googlers” included equations, data tables, and graphs. These showed how much data Google could “read” in a second. When I read these papers, I had just completed a test of read speed on what were in 2004 reasonably zippy servers. These were IBM NetFinity 5500s. Each server had four gigabytes of random access memory, six internal high-speed SCSI drives, IBM Serveraid controllers with on board caching, and an EXP3 storage unit holding 10 SCSI III drives. For $20,000, these puppies were fast, stable, and reliable. My testing revealed that a single NetFinity 5500 server could read 65 megabytes per second. I thought that this was good, not as fast as the Sun Microsystems’ fiber server we were testing but very good.

The Google data reported that using IDE drives identical to the ones available at the local Best Buy or Circuit City, Google engineers reported read speeds of about 600 megabytes per second. Google was using off-the-shelf components, not exotic stuff like IBM Serveraid controllers, IBM-proprietary motherboards, IBM-certified drives, and even IBM FRU (field replaceable unit) cables. Google was using the low cost stuff in my father’s PC.

A Google server comparable to my NetFinity 5500 cost about $600 dollars in 2004. The data left me uncertain of my analysis. So, I had two of my engineers retrace my tests and calculations. No change. I was using a server that cost 33 times as much as Google’s test configuration server. I was running at one-tenth the read speed of Google’s server. One-tenth the speed and spending $19,400 more per server.

You don’t have to know too much about price – performance ratios to grasp the implications of these data. A Google competitor trying to match Google’s “speed” has to spend more money than Google. If Google builds a data center and spends $200 million, a competitor in 2004 using IBM or other branded server-grade hardware would have to spend orders of magnitude more to match Google performance.

The gap in 2004 and 2005 when my study The Google Legacy appeared was so significant as to be unbelievable by those who did not trouble to look at the Google data.

This is just one example of what Google has done to have a competitive advantage. I document others in my 2007 study Google Version 2.0.

In the months after The Google Legacy appeared, it struck me that only Amazon of Google’s Web competitors seemed to have followed a Google-like technical path. Today, even though Amazon is using some pretty interesting engineering short cuts, Amazon is at least in the Google game. I’m still watching the Amazon S3 meltdown to see if Amazon’s engineers have what it takes to keep pace. Amazon’s technology and research budget is a pittance compared to Google’s. Is Amazon able to out-Google Google? It’s too early to call this horse race.

Do Other Companies See Google More Clearly than Microsoft Did?

Now, let me shift to the connection I made between Mr. Stross’s article and the letter I mentioned.

Some disclaimers. This confidential letter was not addressed to me. A colleague allowed me to read the letter. I cannot reveal the name of the letter’s author or the name of my colleague. The letter’s author is a senior executive at a major computer company. (No, the company is not Microsoft.)

My colleague proposed a strategy analysis of Google to this big company. The company sent my colleague a “go away” letter. What I remember are these points: [a] (I am paraphrasing based on my recollection) our company has a great relationship with Google, and we know what Google is doing because our Google contacts are up front with us. [b] Our engineers have analyzed Google technology and found that Google’s engineering poses no challenge to us. [c] Google and our engineers are working together on some projects, so we are in daily contact with Google. The letter concluded by saying (again I paraphrase), “Thanks, but we know Google. Google is not our competitor. Google is our friend. Get lost.”

I had heard similar statements from Microsoft. But when wrapping up Google Version 2.0, I spoke with representatives of Oracle and other companies. What did I hear? Same thing: Google is our partner. Even more interesting to me was that each of these insightful managers told me their companies had figured out Google.

How Did Microsoft Get It Wrong?
This begs the question, “How did Microsoft and its advisors not figure out Google?” Microsoft has known about Google for a decade. Microsoft has responded on occasions to Google’s hiring of certain Microsoft wizards like Kai Fu Lee. Microsoft has made significant commitments to search and retrieval well before the Yahoo deal took shape. Microsoft has built an advanced research capability in information retrieval. Microsoft has invested in an advertising platform. Microsoft has redesigned Microsoft Network (MSN) a couple of times and created its own cloud-computing system for Live CRM, among other applications.

I don’t think Microsoft got it wrong. I think Microsoft looked at Google through the Microsoft “agenda”. Buttressed by the received wisdom about Google, Microsoft did not appreciate that Google’s competitive advantage in ads was deeply rooted in engineering, cost control, and its application platform. Perhaps executives in other sectors may want to step back and ask themselves, “Have we really figured out Google?”

Let’s consider Verizon’s perception of Google.

I want to close by reminding you of the flap over the FCC spectrum bid. The key development in that process was Verizon’s statement that it would become more open. Since the late 1970s, I have worked for a number of telcos, including the pre-break up AT&T, Bell Labs as a vendor, USWest before it became Qwest, and Bell Communications Research. When Verizon used the word open, I knew that Google has wrested an important concession from Verizon. Here’s Business Week’s take on this subject. At that moment, I lost interest in the outcome of the spectrum auction. Google got what it wanted, openness. Google’s nose was in Verizon’s tent. Oh, Verizon executives told me that Google was not an issue for them as recently as June 2007.

What’s happening is far larger than Microsoft “with wobbly legs, scared witless” to quote Mr. Stross. Microsoft, like Verizon, is another of the established, commercial industrial giants to feel Google’s pressure wave. Here’s a figure from The Google Legacy and the BearStearns’ report The Google Ecosystem to illustrate what Googl’s approach was between 1998 and 2004. More current information appears in Google Version 2.0.

Pressure wave

You can figure out some suspects yourself. Let me give you a hint. Google is exerting pressure in its own Googley way on the enterprise market. Google is implementing thrusts using these pressure tactics in publishing, retail, banking, entertainment, and service infrastructure. Who are the top two or three leaders in each of these sectors? These are the organizations most likely to be caught in the Google pressure wave. Some will partner with Google. Others will flee. A few will fight. I do hope these firms know what capabilities Google can bring to bear on them.

The key to understanding Google is setting aside the Web search and ad razzle dazzle. The reality of Google lies in its engineering. Its key strength is its application of very clever math to make tough problems easy to resolve. Remember Google is not a start up. The company has been laboring for a decade to build today’s Google. It’s also instructive to reflect on what Google learned from the former AltaVista.com wizards who contributed much in the 1999 – 2004 period; many continue to fuel Google’s engineering prowess today. Even Google’s relationship with Xooglers (former employees who quit) extends Google’s pressure wave.

I agree that it is easy, obvious, and convenient to pigeon hole Google. Its PR focuses on gourmet lunches, Foosball, and the wacky antics of 20-year-old geniuses. Too few step back and realize that Google is a supra-national enterprise the likes of which has not been experienced for quite a while.

My mantra remains, “Surf on Google.” The alternative is putting the fate of your organization in front of Googzilla and waiting to see what happens. Surfing is a heck of a lot more fun than watching the tsunami rush toward you.

The key findings from this two and a half year effort are two:

  1. Google has morphed into a new type of global computing platform and services firm. The implications of this finding mystify wizards at some very large companies. The perception of Google as a Web search and online ad company is so strongly held that no other view of Google makes sense to these people.
  2. The application platform is more actively leveraged than most observers realize. Part of the problem is that Google is content to be “viral” and low key. Pundits see the FCC spectrum bid as a huge Google initiative. In reality, it’s just one of a a number of equally significant Google thrusts. But pundits “see” the phone activities and make mobile the “next big thing” from Google.

Stephen Arnold, February 25, 2008

Comments

Comments are closed.

  • Archives

  • Recent Posts

  • Meta